You are on page 1of 22

University of Nairobi

College of Health Sciences


Institute of Tropical and Infectious Diseases

WID 705: EPIDEMIOLOGY AND BIOSTATISTICS

Lecture 10: Data Quality Control

Crispin Matere
cmatere@gmail.com

9th March, 2022


© 2022-UNITID: Crispin Matere
Introduction

 As researchers, we know that research results are


only as good as the data from which they were
derived. It is therefore critical that research data
is of the highest quality possible. In this
presentation we examine certain processes and
procedures that we believe are critical for high
quality data, given available data sources,
systems, and personnel.
Data Quality Control

 Data quality means fitness of the data to serve


the purposes for which it was acquired. It is a
multi-dimensional concept and is defined by
various factors such as accuracy, completeness,
consistency, timeliness, security, integrity etc.
 Data quality control refer to those processes and
procedures that are used to produce high quality
data in health research setting
Errors and Bias

 Bias is a tendency to lean in a certain direction,


or a prejudice towards something. Biases lead to
poor quality data.
 Data errors can be defined as deviations from
true values. Types of errors include;
 a) Careless errors: Are due to mistakes in
reading scales or careless setting of markers, etc.
They can be eliminated by repetition of readings
by one or two observers.
Errors and Bias cont..

 b) Systematic errors: Are due to defects in the


data collection instruments either in design or
calibration. They are the hardest source of
errors to detect.
 c) Random errors: Are due to natural
fluctuations of physical parameters or the
judgement of the experimenter, such as variation
in response time or estimation in instrument
reading.
Sources of errors in health research

 In health research we can identify at least five


sources of data errors in which data quality
control efforts should be directed. These
include;
 1. Population Specification
 This type of error occurs when the researcher
selects an inappropriate population or universe
from which to obtain data.
Sources of errors in health research

 In health research we can identify at least five


sources of data errors in which data quality
control efforts should be directed. These
include;
 1. Population Specification
 This type of error occurs when the researcher
selects an inappropriate population or universe
from which to obtain data.
Sources of errors in health research cont..

 3. Selection
 Selection error is the sampling error for a sample
selected by a nonprobability method. In health
researchers may just work with those patients
that access health facility.
 4. Non-response
 Nonresponse error can exist when an obtained
sample differs from the original selected sample.
Often occur when one is unable to reach some
of the respondents.
Sources of errors in health research cont..

 5. Measurement
 Measurement error occurs due the measurement
process itself, and represents the difference
between the information generated and the
information needed by the researcher.
Data Quality Control

 1. Appropriate Research Study Design


 Choose an appropriate study design for your
research based on the following considerations
 Your research objectives
 Types of data to be collected
 Personnel available for data collection
 Time and resources available
Data Quality Control cont..

 2. Data Collection and Management


 Develop a plan for data collection and
management and adhere to it throughout the
research undertaking.
 Define all data you will collect upfront
 Plan your data collection modality by
considering the following;
 What is the structure of the data?
Data Quality Control cont..

 Will the study personnel meet face-to-face with


the subjects?
 Can the data be collected remotely?
 Does the data already exist or does it need to be
collected?
 Who is responsible for what
 How will be the data collected
Data Quality Control cont..

 What tools will be used for data collection


(study instruments)
 Ensure capacity for data collection and
management
 Maintain data security and confidentiality
Data Quality Control cont..

 3. Setting up Standard Operating procedures


(SOPs)
 The complete data collection and management
process should be documented. The importance
of SOPs is underscored by the fact that
documented work procedures are mandated by
the researcher.
Data Quality Control cont..

 4. Data publishing and sharing results


 • Why do we need to share?
 reinforcing open scientific inquiry
 encouraging diversity of analysis and opinion,
 promoting new research, testing of new or
alternative hypotheses and methods of analysis
 supporting studies on data collection methods
and measurement
Data Quality Control cont..

 facilitating training on data issues for new


researchers
 enabling the exploration of topics not
envisioned by the initial investigators
 permitting the creation of new datasets by
combining data from multiple sources
Data Quality Control cont..

 5. Sharing your methods and resources


 It is important for your data management and
analysis to be:
 Automated (when possible)
 Sharable with your peers
 Reproducible by others
 Downloadable as sourcecode or virtual machine
(VMs)
 Reporting and analysis templates
Data Quality Control cont..

 6. Monitoring and evaluation system


 To maintain data quality throughout the data
life-cycle a robust M&E system is inevitable.
Assessing Data Quality

 1.) Data integrity checking


 Questionable data generates data queries.
 Open a data query
 Investigate
 Document response
 Resolve or repeat
Assessing Data Quality

 (2) Source Document Verification


 Can be time consuming, expensive, stressful.
 May capture ten times as many errors as data
integrity checks.
 Process:
 Take the data you’ve received for your research
study. Get access to the source documents for
that data. See if the data values from both
sources match.
Summary

 For data and information collected and managed


in health research strive to reduce errors in
measurement:
 1. develop well standardized protocols
 2. train interviewers and technicians well
 3. perform pilot studies to identify problems
with questionnaires and measuring instruments
 4. carry out a thorough data check and cleaning
 Follow study protocol when carrying out your
research.
References

 WHO, 2003. Improving Data Quality: A Guide


for Developing Countries.
 Microsoft Word - DQM
Manual_FINAL_November 2014.docx
(pactworld.org)
 Data Quality — MEASURE Evaluation
 https://www.coursera.org/learn/clinical-data-
management

You might also like